Synthesia CEO Urges Shift From AI Token Metrics to Outcomes
Victor Riparbelli argues that businesses must measure AI adoption through concrete business outcomes and efficiency rather than the volume of tokens used.
Victor Riparbelli, CEO of Synthesia, is calling for a fundamental shift in how businesses measure the adoption of artificial intelligence. He argues that companies are currently relying on vanity metrics, specifically the volume of AI tokens used, which he describes as tokenmaxxing. Riparbelli compares this trend to the historical reliance on pageviews in newsrooms, stating that both approaches incentivize gaming the system instead of creating long-term value.
To replace these metrics, Riparbelli proposes an AI impact scorecard. This framework focuses on three primary pillars: efficiency, measured by time and cost reduction; quality, measured by behavior change and application; and compliance, measured by auditability and regulatory adherence. He recommends that organizations reward verified impact over volume and measure success at the workflow level rather than the model level.
As an example of this utility-over-novelty approach, Riparbelli points to the use of pre-learning videos by the elevator company Machine Learning. The company utilized AI-generated content to produce materials 30% faster and reduced the time required for each course by one day.